CN100565589C - The apparatus and method that are used for depth perception - Google Patents

Abstract

The invention discloses a method for rendering a multi-viewpoint image comprising a first output image and a second output image based on an input image (102). The method includes: creating a modulated image (100) comprising irregularly shaped objects (106-112); modulating pixel values of a portion of an input image (102) based on other pixel values of the modulated image (100), thereby forming image (104); and generating a multi-viewpoint image by warping the intermediate image based on the disparity data.

Description

Apparatus and method for depth perception

technical field

The invention relates to a method for rendering multi-viewpoint images based on input images and parallax data.

The invention also relates to a rendering unit for rendering multi-viewpoint images based on input images and disparity data.

The invention also relates to an image processing device comprising the rendering unit.

The invention also relates to a computer program product loaded by a computer configuration comprising processing means and memory, comprising instructions for rendering multi-viewpoint images based on input images and disparity data.

Background technique

Since the introduction of display devices, many people have dreamed of realistic 3-D display devices. Many principles by which such display devices can be realized have been investigated. Some principles attempt to create realistic 3-D objects in a certain space. For example, A. Sullivan in the article "Solid-state Multi-planar Volumetric Display" in proceedings of SID'03 (1531-1533, 2003) discloses a display device that moves visual data over a series of planes via a fast projector . Each plane is a switchable diffuser. If the number of planes is large enough, the human brain assembles the pictures and sees realistic 3-D objects. This principle allows the viewer to look around the object within a certain range. In this display device all objects are (semi)transparent.

Many other principles attempt to create 3-D display devices based solely on binocular parallax. In these systems, the viewer's left and right eyes perceive different images, so the viewer perceives a 3-D image. An overview of these concepts can be found in the book "StereoComputer Graphics and Other True 3-D Technologies" (D.F. McAllister (Ed.)), Princeton University Press, 1993. The first principle uses anaglyph glasses in combination with eg a CRT. If an odd frame is displayed, the light is blocked to the left eye, and if an even frame is displayed, the light is blocked to the right eye.

A display device that does not require additional tools to display 3-D is called an auto-stereoscopic display device.

The first glasses-free display devices included gobos to create cones of light aimed at the viewer's left and right eyes. For example, the cones of light correspond to odd and even sub-pixel columns, for example. These columns are addressed with the appropriate information, and if the viewer is at the correct point, different images are obtained in his left and right eyes and a 3-D picture can be perceived.

The second glasses-free display device includes a series of lenses to image light from odd and even sub-pixel columns to the viewer's left and right eyes.

A disadvantage of the glasses-free display devices described above is that the viewer must remain in a fixed position. In order to guide the viewer, it has been proposed to use pointers to show the viewer in the correct position. See for example US Pat. No. 5,986,804, where a visor is combined with red and green LEDs. With the correct orientation of the viewer, he sees green or red light.

In order to free the viewer from being in a fixed position, multi-viewpoint auto-stereoscopic display devices have been proposed. See for example US60064424 and US20000912. In the display devices disclosed in US60064424 and US20000912, inclined lenticular lenses are used, whereby the width of the lenticular lens is larger than two sub-pixels. In this way there are several adjacent images and the viewer can move left and right with some freedom

In order to generate a 3-D impression on a multi-view display device, images must be rendered from different virtual viewpoints. This requires multiple input views or some 3-D or depth information. This depth information can be recorded, generated from a multi-view camera system, or generated from conventional 2-D video material. To generate depth information from 2-D videos, several depth cues can be employed: e.g. motion structure, focus information, geometry, and dynamic occlusion. The goal is to generate a dense depth map, i.e. one depth value per pixel. Next, the depth map is used to render the multi-view image to give the viewer the impression of depth. In the article "Synthesis of multi viewpoint images at non-intermediate positions" by P.A. Redert, E.A. Hendriks and J. Biemond (Proceedings of International Conference on Acoustics, Speech and Signal Processing, Vol. IV, ISBN 0-8186-7919-0, pp. 2749- 2752, IEEE Computer Society, Los Alamitos, California, 1997) discloses a method for extracting depth information and rendering a multi-viewpoint image based on an input image and a depth map. A multi-view image is a set of images to be displayed by a multi-view display device to create a 3-D impression. Typically, the set of images is created based on input images. Create one of these images by shifting the pixels of the input image by the corresponding offset. These offsets are called parallax. Thus, typically, for each pixel there is a corresponding disparity value, which together form a disparity map. Typically, the disparity value is inversely proportional to the depth value, ie:

$\mathrm{<span\; class="notranslate">\; S</span>}<span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; C</span>}}{\mathrm{<span\; class="notranslate">\; D.</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

where S is the disparity, C is a constant and D is the depth. Creating a depth map is considered equivalent to creating a disparity map.

For uniform regions of a 2-D input image, ie substantially texture-free regions, it is difficult or sometimes impossible to infer from a multi-view display device what its depth is. Typically, this will be perceived as an object corresponding to a uniform area at screen level. With a uniform background such as a blue sky, the perceived depth of a multi-view display device is relatively small. In the case of a cloudless sky, the sky is perceived to be at screen level, so for a correct depth impression it is impossible to place other objects behind the screen, which seriously reduces the depth impression.

Contents of the invention

It is an object of the invention to provide a method as described in the opening paragraph, whereby the impression of depth is increased.

Above-mentioned purpose of the present invention is achieved like this, and this method comprises:

- Create modulated images including irregularly shaped objects;

- modulating pixel values of a portion of the input image on the basis of other pixel values of the modulating image, thereby forming an intermediate image; and

- Generation of multi-view images by warping intermediate images on the basis of disparity data.

Giving the viewer a 3-D impression on a multi-view display device depends on showing the first output image to the left eye and the second output image to the right eye. The difference between these output images is interpreted into a 3-D image by the human brain. The output image is constructed by displacing the objects of the input image relative to each other. The offset is determined by the object's depth. The brain recognizes correspondences between objects at different viewpoints, i.e. output images, and infers geometric shapes from the differences. If the object is substantially textureless, it is difficult to make such a correspondence, since there is no feature of eye "lock-in". Imaging a uniform black surface. Shifting it left or right doesn't change it. Therefore, the depth at which the surface is located cannot be inferred based on parallax.

Features are introduced by modulating pixel values of a portion of an input image based on other pixel values of the modulating image. These features corresponding to irregularly shaped objects are first visible in substantially uniform regions of the input image. The first and second output images may then be rendered to appear to differ in regions corresponding to the portion of the input image that was substantially uniform prior to modulation. The user can now make a correspondence between the irregularly shaped objects introduced in each of the first and second output images.

Preferably, the size of the irregularly shaped object is related to the disparity data. For example, the average size of an irregularly shaped object is of the same order as the average value of the disparity data. Assuming that the disparity data includes values in the range 1-15 pixels, the advantage is that the dimensions, ie the height and width of the irregularly shaped object, are substantially in the same range. Preferably, the average diameter of the irregularly shaped object is about 7-8 pixels for an image of 1000*1000 pixels. The average diameter refers to the average distance between two edges.

The modulation of the pixel values of the input image may cover pixels spread over the input image. The modulation preferably covers that part of the input image corresponding to a substantially uniform area. Preferably, the modulation is such that the brightness values of a first part of the pixels of the input image are increased, while the brightness values of a second part of the pixels of the input image are decreased. For example, a first portion of pixels of the input image corresponds to a set of pixels representing the modulated image of an irregularly shaped object, while a second portion of pixels of the input image corresponds to another set of pixels of the modulated image representing the background. Preferably, the average brightness value is not affected by the modulation, ie the average brightness value of the input image and the average brightness value of the intermediate image are substantially equal to each other.

In an embodiment of the method according to the invention, creating a modulated image comprises:

- creating the first image by generating noise;

- filtering the first image using a low pass filter, thereby forming the second image; and

- Dividing the pixels of the second image by thresholding to form a modulated image.

Preferably, the noise is generated by a random noise generator. The characteristics of the low pass filter are preferably correlated with the disparity data in order to create irregularly shaped objects of suitable size. The partitioning is done by marking connected groups of pixels as belonging to the respective irregularly shaped object, while marking other groups of connected pixels as background.

In an embodiment of the method according to the invention the pixel values are modulated based on the disparity data. Preferably, the raising and lowering of the luminance value depends on the local depth value and thus on the local disparity value of the pixel. Preferably, the amount of boosting and/or lowering is higher for objects of the input image that are further away from the viewer.

In an embodiment of the method according to the invention, the modulated image is created based on motion vectors calculated on the basis of the series of input images to which the input image belongs. Suppose the method according to the invention is applied to a series of input images representing motion, for example a series of video images. For example, it corresponds to panning a camera. If each input image of the input image sequence is modulated by the same modulation image and displayed on a multi-view display device, the result may appear to be viewing the output image sequence through a dirty window. To prevent this, preferably each input image is modulated by its own modulation image. The modulation image used to modulate a particular input image may be based on other modulation images that were created for previous input images, ie images preceding the particular input image. Advantageously, said other modulated image is based on shifting the modulated image in one direction for modulating a particular input image, and said other modulated image is related to motion in the scene. Preferably, to obtain said other modulated image, the modulated image used to modulate the particular input image is shifted by a motion vector calculated by analyzing or simulating the motion vector field corresponding to the particular input image .

Another object of the invention is to provide a rendering unit as described in the opening paragraph, whereby the impression of depth is increased.

Above-mentioned purpose of the present invention is achieved like this, and this rendering unit comprises:

- creating means for creating modulated images including objects of irregular shape;

- modulating means for modulating pixel values of a part of the input image on the basis of other pixel values of the modulated image to form an intermediate image; and

- Generation of multi-viewpoint images by warping intermediate images on the basis of disparity data.

Another object of the present invention is to provide an image processing device comprising a rendering unit as described in the opening paragraph, thereby increasing the impression of depth.

The above-mentioned purpose of the present invention is achieved like this, and described rendering unit comprises:

- creating means for creating modulated images including objects of irregular shape;

- modulating means for modulating pixel values of a part of the input image on the basis of other pixel values of the modulating image to form an intermediate image; and

- Generation of multi-viewpoint images by warping intermediate images on the basis of disparity data.

Another object of the present invention is to provide a computer program product as described in the opening paragraph, whereby the impression of depth is increased.

The above object of the present invention is achieved in that the above computer program product, after being loaded, provides the processing device with the ability to perform the following operations:

- Create modulated images including irregularly shaped objects;

- modulating pixel values of a portion of the input image based on other pixel values of the modulating image, thereby forming an intermediate image; and

- Generation of multi-viewpoint images by warping intermediate images on the basis of disparity data.

Modifications and variations of the rendering unit may correspond to modifications and variations of the image processing apparatus, method and computer program product.

Description of drawings

These and other aspects of the rendering unit, image processing device, method and computer program product according to the invention will be apparent and elucidated with reference to the following description of implementations and embodiments and with reference to the accompanying drawings in which:

Figure 1 shows a modulated image, an input image and an intermediate image according to the present invention;

Fig. 2 schematically shows an embodiment of a rendering unit according to the present invention;

Fig. 3 schematically shows a multi-viewpoint image generating unit comprising an embodiment of a rendering unit according to the present invention;

Figure 4 schematically illustrates an embodiment of a modulated image creation device; and

Fig. 5 schematically shows an embodiment of an image processing device according to the invention.

The same reference numerals are used throughout the drawings to refer to similar parts.

Detailed ways

Fig. 1 shows a modulated image 100, an input image 102 and an intermediate image 104 according to the invention. The input image 102 is an image from a video sequence. The modulated image 100 and the input image 102 have the same size, ie comprise the same number of pixels. Then the modulated image 100 is directly used to modulate the input image 102 . For each pixel of the input image 102, there is a corresponding pixel in the modulated image 100, which is directly related to the amount of increase or decrease of the respective brightness value. Alternatively, the modulated image 100 and the input image 102 have different sizes from each other. Modulation of the input image 102 is then performed by applying the modulated image 100 multiple times or only a part of the modulated image 100 . Alternatively, only a fraction of the pixels of the input image are modulated.

Preferably, the modulated image 100 includes a first set of connected pixels 114 and a second set of pixels, wherein the first set of pixels together form the background and the second set of pixels form the foreground objects 106-112. These foreground objects are irregularly shaped objects. These irregularly shaped objects 106-112 appear as blotches. Preferably, the shapes of these irregularly shaped objects 106 - 112 do not correlate with the shapes of the objects in the input image 102 .

The average size of these irregularly shaped objects 106-112 is related to the amount of parallax, and thus depth. Note that different irregularly shaped objects 106-112 may have different sizes from each other. Also, typically, the amount of parallax of different pixels of the input image 102 exhibits a deviation, and thus the amount of parallax of the intermediate image 104 also exhibits a deviation. However, the average size of the parallax and the average size of the irregularly shaped objects 106-112 are preferably of the same order of magnitude.

Figure 1 shows an intermediate image 104 according to the invention. Irregularly shaped objects 106-112 are clearly visible. Note that the intermediate image 104 shown is only an example to illustrate the exaggerated modulation effect. Preferably, irregularly shaped objects are less perceptible. That means they shouldn't be so obvious. Typically, the range and number of apparent luminance values in modulated image 100 are relatively small compared to the number of luminance values in input image 102 . Assume that the range of luminance values of the input image 102 includes 256 different values. The range of luminance values of the modulated image 100 then typically comprises the values [-2, 2]. For example, the brightness values of the first group of pixels, i.e. the brightness values of the background 114, are all equal to -2 or -1, while the brightness values of the second group of pixels, i.e. the brightness values of the irregularly shaped objects 106-112, are all equal to +2 or +1.

Fig. 2 schematically shows an embodiment of a rendering unit 200 according to the present invention. The rendering unit 200 is configured to render a multi-view image including a first output image and a second output image based on the input image 102 . The input image 102 is provided at an image input connector 208 . The rendering unit 200 provides a first output image and a second output image at its image output connectors 210 and 212 . The rendering unit 200 includes:

- modulation image creation means 206 for creating a modulation image 100 comprising irregularly shaped objects 106-112;

- modulating means 202 for modulating pixel values of a part of the input image 102 on the basis of other pixel values of the modulated image 100, thereby forming an intermediate image 104; and

- generating means 204 for generating a first output image and a second output image wherein the first output image is generated by distorting the intermediate image on the basis of a first transformation based on disparity data and the second output image is generated by Generated by distorting the intermediate image based on the second transformation of the disparity data.

Modulated image creating means 206, modulating means 202 and generating means 204 may be implemented using one processor. Typically, these functions are performed under the control of a software program product. During execution, the software program product is typically loaded into a memory, such as RAM, and executed from there. The program may be loaded from a background memory such as ROM, hard disk or magnetic and/or optical storage, or via a network such as the Internet. Optionally, an ASIC provides the functions.

An embodiment of the modulated image creating means 206 is described with reference to FIG. 4 .

Preferably, the modulation means 202 is arranged to perform the function as specified in Equation 2.

L _out (x, y)=L _in (x, y)+g(x, y)*L _mod (x, y) (2)

in,

-L _in (x, y) is the brightness value of the pixel whose coordinates are (x, y) of the input image 102;

- L _out (x, y) is the brightness value of the pixel with coordinates (x, y) of the intermediate image 104, i.e. the output of the modulation means;

- L _mod (x, y) is the brightness value of the pixel at coordinates (x, y) of the modulated image 100; and

-g(x,y) is the gain factor, preferably it can be adjusted by the user. The gain g(x,y) may be equal for all pixels, but preferably each pixel has its own gain factor. The actual value of the gain g(x, y) can be provided through the gain input connector 214 .

The generating means 204 is arranged to render the first output image and the second output image. For example, the rendering is as described in the article "Synthesis of multi viewpoint images at non-intermediate positions" by P.A. Redert, E.A. Hendriks and J. Biemond (Proceedings of International Conference on Acoustics, Speech and Signal Processing, Vol. IV, ISBN 0-8186-7919 -0, pp. 2749-2752, IEEE Computer Society, Los Alamitos, California, 1997). Alternatively, the rendering is as described in the article "High-quality images from 2.5D video" by R.P. Berretty and F.E. Ernst (Proceedings Eurographics, Granada, 2003, Short Note 124). For this rendering, the generation means 204 requires disparity or depth information provided by the disparity input connector 216 .

The modulated image creation device 206 may include the following two optional input connectors: a resolution input connector 220 and a motion vector input connector 218 .

Preferably, the introduction of irregularly shaped objects in the input image is limited to that portion of the input image that is substantially uniform. This can be achieved by modulating the input image only locally, ie in a substantially uniform area. Alternatively, the modulated image creation means 206 takes into account information about the image content of the input image, in particular the presence and location of homogeneous regions. This information may be provided by an external resolution calculation device 302, or may be calculated by the rendering unit 200 itself. In both cases, the sharpness information is determined based on calculations of sharpness values for image pixels. Preferably, it is a specific input image into which the modulated image can be added, or with which the modulated image can be merged. Alternatively, the sharpness information is determined based on calculation of sharpness values of pixels of another image from the image sequence to which the particular input image belongs.

Preferably, the sharpness value of a specific pixel is determined by calculating the difference between the brightness and/or color value of the specific pixel and the brightness and/or color values of neighboring pixels of the specific pixel. A sharpness map is formed by calculating the sharpness value of each pixel of the image. Relatively large differences between brightness and/or color values imply relatively high sharpness values. Next, analyze and optionally modify the sharpness map. This means that a first area with more pixels with lower sharpness values is determined and a second area with more pixels with higher sharpness values is determined. It is assumed that the first region is a uniform region, and the second region is a texture region or a detailed region (detailed region). Based on this division, the value of the gain factor g(x,y) is determined and the modulation image 100 is created. Typically, this means that the luminance values L _mod (x, y) of the modulated image 100 corresponding to the first region are such that they have no or substantially no influence on the input image 100 during modulation, e.g. L _mod (x , y)=0, while the luminance values L _mod (x, y) of the modulated image 100 corresponding to the second region are such that they have an influence on the input image 100 during modulation, for example Lmod (x, y)= -2, -1, 1 or 2.

The resolution map including division information is supplied to the rendering unit 200 through the resolution input connector 220 .

The creation of subsequent modulated images, corresponding to subsequent input images, can be done completely independently of each other. Alternatively, there is a link between the creation of a particular modulated image and subsequent modulated images. It is beneficial to account for motion between subsequent input images by creating subsequent modulated images. Shifts can be determined by analyzing the motion between a particular input image and its successors. Preferably, the shift is applied to shift a particular modulated image in order to obtain the next modulated image. Preferably, the motion between subsequent input images depends on establishing a motion model based on the motion vector field. This motion vector field is determined by a motion estimator. This motion estimator can be found, for example, in the article "True-Motion Estimation with 3-D Recursive Search Block Matching" by G. de Haan et al. (IEEE Transactions on circuits and systems for video technology, vol.3, no.5, October 1993 , pp. 368-379).

The motion information is provided to the rendering unit 200 through the motion vector input connector 218 .

Fig. 3 schematically shows a multi-viewpoint image generation unit 300 including an embodiment of the rendering unit 200 according to the present invention. The multi-viewpoint image generating unit 300 is configured to generate a series of multi-viewpoint images based on a series of video images. The multi-view image generation unit 300 has a video image stream at input connector 308 and provides two related video image streams at output connectors 310 and 312, respectively. The two related video image streams are provided to a multi-view display device for visualizing a first series of views based on a first of the related video image streams and for visualizing a first series of views based on the first of the related video image streams. A second series of view visualizations for a second video image stream in the related video image stream. If the user, ie the viewer, observes the first series of views with his left eye and the second series of views with his right eye, he gets a 3-D impression. The first video image stream in the associated video image stream may correspond to the received video image sequence, and the method according to the present invention may evaluate the second video image in the associated video image stream based on the received video image sequence. The image stream is rendered. Preferably, two video image streams are rendered according to the method of the present invention based on the received video image sequence.

The multi-viewpoint image generating unit 300 also includes:

- Sharpness calculation means 302 for determining which regions of the input image are homogeneous. The output of the sharpness calculation means 302 is provided to the rendering unit 200 through the sharpness input connector 220 .

- A motion estimator 304 for estimating motion between subsequent input images. providing the output of motion estimator 304 to rendering unit 200 via motion vector input connector 218; and

- A depth creation unit 306, configured to determine the depth information of each object in the input image. Based on this depth information a disparity map is determined which is supplied to the rendering unit 300 via the disparity input connector 216 .

Note that although the multi-viewpoint image generation unit 300 is designed to process video images, alternative embodiments of the multi-viewpoint image generation unit 300 may also be configured to generate multi-viewpoint images based on a single image, ie, a still picture.

Note that although the multi-viewpoint image generation unit 300 has two output connectors 310 and 312, alternative output methods are also possible. Besides, the number of output images forming one multi-viewpoint image is not strictly limited to two.

Fig. 4 schematically shows an embodiment of a modulated image creation device 206 according to the present invention. The modulation image creation device includes:

- a random noise generator 402 for creating the first image;

- A low pass filter 404 for filtering the first image to form the second image. The characteristics of the low-pass filter are correlated with the disparity data in order to create irregularly shaped objects of suitable size; and

- comparing means 406 for comparing the pixel values of the second image with predetermined thresholds in order to divide the pixels of the second image to form a modulated image. The division is made such that groups of connected pixels are marked as belonging to each irregularly shaped object, while other groups of connected pixels are marked as background.

Fig. 5 schematically shows an embodiment of an image processing device 500 according to the present invention, including:

- a receiving unit 502 for receiving a video signal representing an input image;

- a multi-viewpoint image generating unit 300, configured to generate a multi-viewpoint image based on the received input image, as described in connection with FIG. 3; and

- a multi-viewpoint display device 504 for displaying the multi-viewpoint images provided by the multi-viewpoint image generation unit 300 .

The video signal may be a broadcast signal received via an antenna or cable, but may also be a signal from a storage device such as a VCR (Video Recorder) or a Digital Versatile Disc (DVD). The signal is provided at input connector 506 . The image processing device 500 may be, for example, a TV. Alternatively, the image processing device 500 does not include an optional display device, but provides an output image to a device including the display device 504 . The image processing device 500 may then be eg a set top box, satellite tuner, VCR player, DVD player or recorder. Optionally, the image processing device 500 includes a storage device such as a hard disk or a device for storage on a removable medium such as an optical disc. The image processing device 500 may also be a system applied by a movie company or a broadcaster.

It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design alternative embodiments without departing from the scope of the claims. In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. The word "comprising" does not exclude elements or steps not listed in a claim. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several precise elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means can be embodied by one and the same item of hardware. The use of the words first, second and third etc. does not indicate any ordering. These words can be interpreted as the same.

Claims (9)

1. A method for rendering a multi-viewpoint image based on an input image (102) and parallax data, comprising: - creating a modulated image (100) comprising irregularly shaped objects (106-112); - modulating pixel values of a portion of the input image (102) on the basis of corresponding pixel values of the modulating image (100), thereby forming an intermediate image (104); and - Generating a multi-view image by warping the intermediate image (104) on the basis of the disparity data. 2. The method of claim 1, the size of the irregularly shaped object (106-112) being related to the disparity data. 3. The method of claim 2, the mean size of the irregularly shaped objects (106-112) and the mean value of the disparity data are of the same order of magnitude. 4. A method as claimed in any preceding claim, the part of the input image (102) being a uniform area. 5. The method of claim 4, creating the modulated image (100) comprising: - creating the first image by generating noise; - filtering the first image using a low pass filter, thereby forming the second image; and - Dividing the pixels of the second image by thresholding to form a modulated image (100). 6. The method of claim 4, modulating pixel values based on disparity data. 7. The method of claim 4, creating the modulated image (100) based on motion vectors calculated based on the sequence of input images (102) to which the input image belongs. 8. A rendering unit (200) for rendering a multi-viewpoint image based on an input image (102) and disparity data, the rendering unit comprising: - creating means (206) for creating a modulated image (100) comprising irregularly shaped objects (106-112); - modulating means (202) for modulating pixel values of a part of the input image on the basis of corresponding pixel values of the modulating image (100) to form an intermediate image (104); and - Generating means (204) for generating a multi-viewpoint image by warping the intermediate image (104) on the basis of disparity data. 9. An image processing device (500), comprising: - receiving means (502) for receiving a signal corresponding to the input image (100); - a rendering unit (200) for rendering multi-viewpoint images according to claim 8; and - Display means (504) for displaying multi-viewpoint images.

Images